Marshaling polling data and statistics, The Huffington Post's Ryan Grim makes the case that marijuana users (or at least legalization advocates) could be the key to the 2010 midterms. Why? Because local and statewide pot initiatives could pull liberal voters to the polls in a similar way that gay marriage referendums spurred "values voters" in 2004. The key to Grim's theory is the "surge voter," described as a person who voted because of "a once-in-a-generation mix of shame at the outgoing administration and hope in a new, barrier-breaking candidate." They don't make up a huge demographic percentage, but those few extra points could sway some closely contested races. Here's a snapshot of some of them.

In California:

Because Democrats have yet to coalesce around reform of marijuana laws, the effort to link pot to the party's electoral hopes is going on quietly, the opposite of Rove's coordinated campaign with religious groups opposed to marriage rights for all Americans. A recent survey has California's marijuana initiative up 56-42, though proponents worry about the threat of major spending in opposition from the prison guards union, alcohol interests and pharmaceutical companies.
In Nevada:
Nevada, a swing state, has twice rejected pot legalization initiatives in the past, though support increased to 44 percent in 2004, the last time it was on the ballot. Supporters plan to put it on again in 2012. Whether it can pass isn't some Democrats' top concern: As long as it can get unlikely voters to a polling station they'd otherwise avoid, it's a success.
In Colorado:
...Surge voters, single women under 40 and Hispanics all told America Votes pollsters that if a legalization measure were on the Colorado ballot, they'd be more likely to come out to vote. Forty-five percent of surge voters and 47 percent of single women said they'd be more interested in voting if the question was on the ballot.
UPDATE: The Atlantic's Joshua Green, who weighed in on this topic several weeks ago, counters Grim's article with this claim: "Political scientists disagree about whether Rove's [2004] initiatives helped Republicans. That's because ballot initiatives turn out opponents as well as supporters, which mitigates their effect."